Open AstroRobin opened 3 years ago
Adding weights to the standard 2D histograms is easy (and has been added as of v0.5.11.0), but for the generalised version is a harder problem. I haven't found method that does the same as binned_statistic_2d
, so for the moment the only option I can think of would be to re-invent the wheel and make our own version. I'm not convinced that we can easily make it as fast as scipy's implementation. Furthermore, I think I understand why the might have ignored the problem, as a several "weighted statistics" are undefined (e.g., what's a weighted maximum?).
I will leave for the time being this issue open, with the understanding that our solution for the moment is not complete.
It looks like there is no easy way to parse weights into the hist2d() function. As kwargs are given to pcolormesh(), we will need to create a parameter "weights" which can then be given to histogram_2d() or binned_statistics_2d().
There probably does not need to be many checks on this parameter as most of the exceptions will be raised by numpy/scipy.